ISSI Institute of Control & Computation Engineering


The main research activity in the area of soft computing is concentrated on neural and neuro-fuzzy networks, evolutionary computation, and their integration. Theoretical investigations of neural networks are focused on: multilayer perceptrons with dynamic models, learning algorithms, structure optimization by evolutionary algorithms; GMDH networks and their extensions: multi-output and dynamic networks; learning strategies for ensembles of feedforward networks. Moreover, fuzzy and neuro-fuzzy systems, especially methods of structure design and parameter optimization, are investigated. The elaborated methods of artificial intelligence (AI) are applied to process modelling, identification and pattern recognition. The main emphasis is put on fault detection and diagnosis systems.
Theoretical and simulation analysis in the area of evolutionary computations deals with effectiveness of evolutionary algorithms in finding global optima for strongly non-linear, multi-modal and multi-dimensional objective functions, and their adaptation efficiency in non-stationary environments. The main research topics include phenotype evolutionary algorithms with mutations based on non-isotropic and isotropic α-stable distributions.